Knowledge Hypergraph-Based Multidimensional Analysis for Natural Language Queries: Application to Medical Data

Knowledge Hypergraph-Based Multidimensional Analysis for Natural Language Queries: Application to Medical Data

연구 분야: Databases

학회: International Conference on Computational Science

In recent years, data is continuously evolving not only in volume but also in types and sources, which makes the multidimensional analysis and decision making using traditional approaches a complex and difficult task. In this paper, we propose a three-layer-based architecture to perform multidimensional analysis of natural language queries on health data: 1/ Treatment layer aiming at xR2RML mappings generation and knowledge hypergraph building; 2/ Storage layer allowing mainly to store the RDF triples returned by the query of NoSQL databases, and 3/ Semantic layer, based on a domain ontology which constitutes the knowledge base for the generation of the mappings and the building of the knowledge hypergraph. The originality of our proposal lies in the knowledge hypergraph and its capacity to support multidimensional queries. A prototype is developed and the experiments have shown the relevance of the returned multidimensional query results as well as an improvement over traditional approaches.